Developing a risk score for pancreatic cancer diagnosis using machine learning techniques applied to linked routine data: a pilot study

Ninety four percent of pancreatic cancer patients die within 5 years of diagnosis. This is largely because patients tend not to experience obvious symptoms until the tumour is well advanced and there are no current means of screening. It is now possible, via a blood test, to detect pancreatic cancer where no symptoms are present. However, this test would only be cost effective if given to patients at an increased risk of having pancreatic cancer.

In this project we will use historic GP and hospital data to find out if it’s possible to identify patients who are most likely to have early stage pancreatic cancer. We will compare 3,115 people who were diagnosed with pancreatic cancer to a similar group of patients who were not. Using machine learning techniques, we will examine, for the first time, if there are combinations of particular health problems, illnesses, or symptoms experienced only by patients who are later diagnosed. Ultimately, improving the triage of these patients means that targeted diagnostic tests could lead to the pancreatic cancer being diagnosed earlier and treated more effectively.

avatar

Dr Laura Woods

Assistant Professor in Epidemiology

Laura completed her undergraduate degree in Human Sciences at Oxford University in 1999 and her Master’s degree in Medical Demography in 2001. She joined the Cancer Survival Group in September 2002 where she completed her PhD “International differences in breast cancer survival and ‘cure’ by social deprivation: a comparative study of England and Australia” in September 2006.

avatar

Prof Bernard Rachet

Professor of Cancer Epidemiology

Bernard qualified in medicine in France and worked as a clinician before entering epidemiological research. He completed a PhD in Epidemiology at the International Agency for Research on Cancer (IARC), Lyon, France. He joined LSHTM in 2002 and is currently principal investigator of a Cancer Research UK programme grant, leading a wide range of projects to quantify, describe and explain patterns and trends in cancer survival by socio-economic group, geographic area and ethnicity, as well as extending the methodology and tools for survival analysis, in collaboration with many research partners in the UK and around the world.

Comments are closed.